PUBLIC OCR SIGN AGE RECOGNITION WITH SKEW & SLANT CORRECTION FOR VISUALLY IMP AIRED PEOPLE

This paper presents an OCR hybrid recognition model for the Visually Impaired People
(VIP). The VIP often encounters problems navigating around independently because they are
blind or have poor vision. They are always being discriminated due to their limitation which can
lead to depression to the VIP. Thus, they require an efficient technological assistance to help
them in their daily activity. The objective of this paper is to propose a hybrid model for Optical
Character Recognition (OCR) to detect and correct skewed and slanted character of public
signage. The proposed hybrid model should be able to integrate with speech synthesizer for VIP
signage recognition. The proposed hybrid model will capture an image of a public signage to be
converted into machine readable text in a text file. The text will then be read by a speech
synthesizer and translated to voice as the output. In the paper, hybrid model which consist of
Canny Method, Hough Transformation and Shearing Transformation are used to detect and
correct skewed and slanted images. An experiment was conducted to test the hybrid model
performance on 5 blind folded subjects. The OCR hybrid recognition model has successfully
achieved a Recognition Rate (RR) of 82. 7%. This concept of public signage recognition is being
proven by the proposed hybrid model which integrates OCR and speech synthesizer.